Automated Machine Learning Algorithm Research Articles

An automated multi-layer perceptron discriminative neural network based on Bayesian optimization achieves high-precision one-source single-snapshot direction-of-arrival estimation

This paper proposes an innovative global solution which is a pioneering work applying automated machine learning algorithms to remarkable precision sparse underwater direction-of-arrival (DOA) estimation that views the subaquatic sparse-sampling DOA estimation problem as a classification prediction task. The proposed solution, termed automated multi-layer perceptron discriminative neural network (AutoMPDNN), is built upon a Bayesian optimization framework. AutoMPDNN transforms sparsely sampled time-domain signals into the complex domain, preserving essential components in a one-source single-snapshot scenario. Leveraging Bayesian optimization principles, the algorithm embeds necessary hyperparameters into the loss function, effectively defining it as a maximum likelihood problem using the upper confidence bound function and incorporating sparse signal features. We also explore the model space architecture and introduce variants of AutoMPDNN, denoted as AutoMPDNNs_ln (n = 2,3,4). Through a series of plane wave simulation experiments, it is demonstrated that AutoMPDNN achieves the highest prediction performance for one-source single-snapshot scenarios compared to classical DOA estimation algorithms that incorporate sparse representation approaches, as well as contemporary deep learning DOA methods under varying conditions.

Machine learning prediction of factors affecting Major League Baseball (MLB) game attendance: algorithm comparisons and macroeconomic factor of unemployment

PurposeThis study aims to identify an automated machine learning algorithm with high accuracy that sport practitioners can use to identify the specific factors for predicting Major League Baseball (MLB) attendance. Furthermore, by predicting spectators for each league (American League and National League) and division in MLB, the authors will identify the specific factors that increase accuracy, discuss them and provide implications for marketing strategies for academics and practitioners in sport.Design/methodology/approachThis study used six years of daily MLB game data (2014–2019). All data were collected as predictors, such as game performance, weather and unemployment rate. Also, the attendance rate was obtained as an observation variable. The Random Forest, Lasso regression models and XGBoost were used to build the prediction model, and the analysis was conducted using Python 3.7.FindingsThe RMSE value was 0.14, and the R2 was 0.62 as a consequence of fine-tuning the tuning parameters of the XGBoost model, which had the best performance in forecasting the attendance rate. The most influential variables in the model are “Rank” of 0.247 and “Day of the week”, “Home team” and “Day/Night game” were shown as influential variables in order. The result was shown that the “Unemployment rate”, as a macroeconomic factor, has a value of 0.06 and weather factors were a total value of 0.147.Originality/valueThis research highlights unemployment rate as a determinant affecting MLB game attendance rates. Beyond contextual elements such as climate, the findings of this study underscore the significance of economic factors, particularly unemployment rates, necessitating further investigation into these factors to gain a more comprehensive understanding of game attendance.

Analysis of gap filling techniques for GRACE/GRACE-FO terrestrial water storage anomalies in Canada

Terrestrial water storage anomaly (TWSA) data derived from the Gravity Recovery and Climate Experiment (GRACE/GRACE-FO) have become indispensable for hydrological monitoring since the first mission launched in 2002. Evaluation of basin dynamics is enabled through TWSA integration with independent hydrological and climate datasets such as precipitation, evapotranspiration, and surface runoff. With over 20 years of observations, long-term statistical analysis reveals water storage trends, provided the 11-month gap between the two missions is filled. This study compares four methods for reconstructing GRACE-like TWSA over Canada, namely: extreme gradient boosting, artificial neural networks, automated machine learning, and projection onto convex sets. The GRACE mascon product released by the Jet Propulsion Laboratory is used over the study region. Nine sets of hydrological and climate parameters are used as predictors for the machine learning models, namely GRACE average seasonal signals, TWSA from the GLDAS Catchment Land Surface Model, precipitation, air temperature, glacier surface mass balance, ocean tides, the North Atlantic Oscillation, the Multivariate ENSO Index, and sea surface temperature. For each of the four methods, 20 % of the GRACE TWSA data (the ‘testing data’) is omitted from the model training, and the model is used to fill the artificial gap. Root mean square error (RMSE) is computed using the difference between testing data and model predictions. Normalized RMSE expresses the RMSE as a proportion of the range of the GRACE TWSA data. The automated machine learning algorithm performs best in terms of mean normalized root mean square error, with 6 % (2.4 cm equivalent water height RMSE), followed by projection onto convex sets, extreme gradient boosting, and artificial neural networks. Inclusion of glacier surface mass balance models derived from the GMAO Modern-Era Retrospective Analysis for Research and Applications Version 2 reanalysis and the Randolph Glacier Inventory improved the automated machine learning average root mean square error by 70 %. Results indicate that filling the gap between missions allows for more comprehensive analysis of terrestrial water storage changes, including basin-by-basin analysis, which is ongoing.

Automated Machine Learning in Bankruptcy Prediction of Manufacturing Companies

Industry 4.0 uses artificial intelligence and machine learning algorithms to optimize processes. The main goal of the manuscript is to analyze the possibilities of applying automated machine learning to predict company bankruptcy. The data sample consists of financial data of 9,771 manufacturing companies for the years 2020 and 2021 collected from the Finstat database. Two methods of automated machine learning, AutoML and H2O, were tested. The results were compared with five other methods - linear discriminant analysis, logistic regression, naive Bayes classifier, CatBoost and XGBoost. The resulting model was cross-validated through the 10-fold approach. The best results were achieved by H2O automated machine learning algorithm with an AUC of 90.13%, followed by the gradient boosting methods CatBoost with AUC of 90.05% and XGBoost (AUC of 88.61%) and another automated learning algorithm AutoML with AUC of 81.17%. The findings of this paper indicate possibilities to apply automated machine learning methods in predicting bankruptcy. However, it is necessary to distinguish between individual automated machine learning algorithms since they provide a different range of results.

Automated Machine Learning Algorithm Using Recurrent Neural Network to Perform Long-Term Time Series Forecasting

Automated Machine Learning Algorithm Using Recurrent Neural Network to Perform Long-Term Time Series Forecasting

자동화 머신러닝(AutoML)과 SHAP을 적용한 청소년의 다중시민성 유형 및 설명 변인 탐색

This study aims to investigate the typology of multiple citizenship among Korean adolescents and the predictive factors in the global era. The study analyzed data from the 6th wave of the 2013 Korean Educational Longitudinal Study and utilized latent profile analysis to explore the typology of multiple citizenship. To identify the predictive factors determining multiple citizenship, an automated machine learning algorithm was used to find the most performant model. SHAP (Shapley additive explanations) techniques were then applied to provide interpretable explanations of the machine learning results. The key findings are as follows. First, in the latent profile analysis, the multiple citizenship of adolescents was categorized into “high multiple citizenship,” “moderate multiple citizenship,” and “low multiple citizenship.” Second, the results of automated machine learning showed that gradient boosting exhibited the highest performance among machine learning techniques. When predicting the key variables related to adolescents’ multiple citizenship through machine learning, it was found that communication competence emerged as the most important predictor, followed by cognitive information processing skills, awareness of cultural and artistic activities, independent career maturity, educational experiences involving cultural diversity and international issues, gender, and relationships with teachers and peers. Based on these results, this study provides policy recommendations and implications for effective civic education for adolescents.

Prolonged vase life by salicylic acid treatment and prediction of vase life using petal color senescence of cut lisianthus

This study investigated the effects of exogenous salicylic acid (SA) concentrations (0, 0.1, 0.3, or 0.5 mM), treatment timing (vegetative or reproductive period), and cultivation methods (soil or hydroponic cultivation) on the vase life of four lisianthus cultivars after flower cutting. A 0.5 mM concentration of the SA application during the reproductive period resulted in the longest average vase life of 15.3 d for the Blue Picote cultivar after flower cutting. A colorimeter and a chlorophyll meter were used to observe and quantify petal color changes after cutting, and the color changes during the senescence stages are visualized using a three-dimensional color space. The three-dimensional visualization demonstrated the color senescence for each measurement period. Based on these data, we used automated machine learning algorithms to predict the vase longevity, and the predictive model showed a good accuracy. This model is expected to be widely used in the floricultural industry.

Fully automated assessment of global longitudinal strain by machine learning predicts death in patients undergoing stress CMR

Abstract Background Several studies described the independent prognostic value of left ventricular global longitudinal strain (GLS) using cardiovascular magnetic resonance (CMR) to predict death and cardiovascular events. While routine GLS evaluation is a time consuming and partially operator-dependent process, the potential interest of GLS using a fully automated method has not yet been established. Purpose To assess the feasibility and prognostic value of GLS to predict all-cause death using a fully automated machine-learning algorithm without human correction in consecutive patients referred for stress CMR. Methods Between 2016 and 2018, all consecutive patients referred for stress CMR were included and followed for the occurrence of all-cause death. A fully automated machine learning algorithm was trained and validated on unseen CMR studies (MAGNETOM Aera and Skyra) to assess the GLS from long-axis cine images acquired at rest. The algorithm combines multiple deep learning networks for detection and segmentation with an active contours approach. Cox regressions were performed to determine the prognostic value of GLS. Results In a retrospective study of 9,883 consecutive patients who underwent stress CMR between 2016 and 2018, automated GLS was successfully computed in 9,638 (97.5%) patients (67% male, mean age 66±12 years). A total of 510 (5.3%) deaths were observed during a median follow-up period of 4.5 years (interquartile range: 3.7-5.2). GLS, the presence of inducible ischemia and late gadolinium enhancement (LGE) were significantly associated with the occurrence of death (hazard ratio, HR: 1.22 [95% CI, 1.17-1.26]; HR: 2.23 [95% CI, 1.61-3.10]; and HR: 2.04 [95% CI, 1.41-2.95], respectively, all p<0.001). After adjustment for traditional risk factors, inducible ischemia and LGE, GLS was an independent predictor of a higher incidence of death (adjusted HR: 1.14 [95% CI, 1.08-1.20]). Automated GLS showed an incremental prognostic value to predict death compared to traditional risk factors, inducible ischemia and LGE (C-statistic improvement: 0.05; NRI=0.146; IDI=0.244; all p<0.001). Conclusions Fully automated GLS measured at rest has an incremental prognostic value to predict all-cause death above traditional risk factors, and other stress CMR parameters. This finding suggests that unsupervised GLS measurements are accurate and allow improved risk stratification.

Identifying leather type and authenticity by optical coherence tomography

PurposeThis study aims to identify leather type and authenticity through optical coherence tomography.Design/methodology/approachOptical coherence tomography images taken from genuine and faux leather samples were used to create an image dataset, and automated machine learning algorithms were also used to distinguish leather types.FindingsThe optical coherence tomography scan results in a different image based on leather type. This information was used to determine the leather type correctly by optical coherence tomography and automatic machine learning algorithms. Please note that this system also recognized whether the leather was genuine or synthetic. Hence, this demonstrates that optical coherence tomography and automatic machine learning can be used to distinguish leather type and determine whether it is genuine.Originality/valueFor the first time to the best of the authors' knowledge, spectral-domain optical coherence tomography and automated machine learning algorithms were applied to identify leather authenticity in a noncontact and non-invasive manner. Since this model runs online, it can readily be employed in automated quality monitoring systems in the leather industry. With recent technological progress, optical coherence tomography combined with automated machine learning algorithms will be used more frequently in automatic authentication and identification systems.

Abstract 268: Validation of an automated method to assess 24‐hour infarct volume in stroke patients

Introduction In stroke patients, accurate infarct volume assessment at 24 hours typically requires manual segmentation of lesions that often are not well defined. We aimed to validate an automated machine learning algorithm (MethinksFIV) specifically trained to automatically segment and measure subacute infarcts on non‐contrast CT (NCCT) scans. Methods We retrospectively studied stroke patients with a large vessel occlusion prospectively admitted to a comprehensive stroke center. The final infarct volume was segmented on the 24h NCCT manually (man‐FIV) and with the automated method (auto‐FIV). Both measurements were correlated with clinical outcomes. Results We included 346 patients. On 24h‐NCCT median auto‐FIV was lower than man‐FIV (8.6 ml [2.3‐40.7] vs. 16.9 ml [1.1‐70.8], p<0.001). Auto‐FIV and man‐FIV were highly correlated (r=0.8; p<0.001; mean dice coefficient of 0.61). The Auto‐FIV correlated better than man‐FIV with 24h (r=0.54 vs. r=0.51, both p<0.001) and discharge NIHSS (r=0.59 vs r=0.54, both p<0.001). An increase of 1 ml in auto‐FIV was associated with an increase in the odds of higher ordinal mRS at three months (OR=1.01 95%CI 1.01‐1.02, p<0.001). Auto‐FIV was a better predictor of mRS>2 at three months (OR: 1.02 95%CI 1.01‐1.13, p=0.002) than manual segmentation (OR=1.01, 95%CI 1.00‐1.01, p=0.01). In a multivariable analysis, only auto‐FIV remained a significant predictor of mRS>2 at three months (aOR: 1.03 95% CI 1.01‐1.05, p=0.004). Conclusion An automated method to measure subacute infarct volumes in stroke patients is accurate and correlates well with further clinical outcomes. The method appears to outperform manual segmentation and may be used to facilitate infarct characterization in large clinical trials.

A Real-Time Automated Machine Learning Algorithm for Predicting Mortality in Trauma Patients: Survey Says it's Ready for Prime-Time.

Though artificial intelligence ("AI") has been increasingly applied to patient care, many of these predictive models are retrospective and not readily available for real-time decision-making. This survey-based study aims to evaluate implementation of a new, validated mortality risk calculator (Parkland Trauma Index of Mortality, "PTIM") embedded in our electronic healthrecord ("EHR") that calculates hourly predictions of mortality with high sensitivity and specificity. This is a prospective, survey-based study performed at a level 1 trauma center. An anonymous survey was sent to surgical providers and regarding PTIM implementation. The PTIM score evaluates 23 variables including Glasgow Coma Score (GCS), vital signs, and laboratory data. Of the 40 completed surveys, 35 reported using PTIM in decision-making. Prior to reviewing PTIM, providers identified perceived top 3 predictors of mortality, including GCS (22/38, 58%), age (18/35, 47%), and maximum heart rate (17/35, 45%). Most providers reported the PTIM assisted their treatment decisions (27/35, 77%) and timing of operative intervention (23/35, 66%). Many providers agreed that PTIM integrated into rounds and patient assessment (22/36, 61%) and that it improved efficiency in assessing patients' potential mortality (21/36, 58%). Artificial intelligence algorithms are mostly retrospective and lag in real-time prediction of mortality. To our knowledge, this is the first real-time, automated algorithm predicting mortality in trauma patients. In this small survey-based study, we found PTIM assists in decision-making, timing of intervention, and improves accuracy in assessing mortality. Next steps include evaluating the short- and long-term impact on patient outcomes.

Validation of an automated machine learning algorithm for the detection and analysis of cerebral aneurysms.

Machine learning algorithms have shown groundbreaking results in neuroimaging. The authors herein evaluated the performance of a newly developed convolutional neural network (CNN) to detect and analyze intracranial aneurysms (IAs) on CTA. Consecutive patients with CTA studies between January 2015 and July 2021 at a single center were identified. The ground truth determination of cerebral aneurysm presence or absence was made from the neuroradiology report. The primary outcome was the performance of the CNN in detecting IAs in an external validation set, measured using area under the receiver operating characteristic curve statistics. Secondary outcomes included accuracy for location and size measurement. The independent validation imaging data set consisted of 400 patients with CTA studies, median age 40 years (IQR 34 years) and 141 (35.3%) of whom were male; 193 patients (48.3%) had a diagnosis of IA on neuroradiologist evaluation. The median maximum IA diameter was 3.7 mm (IQR 2.5 mm). In the independent validation imaging data set, the CNN performed well with 93.8% sensitivity (95% CI 0.87-0.98), 94.2% specificity (95% CI 0.90-0.97), and a positive predictive value of 88.2% (95% CI 0.80-0.94) in the subgroup with an IA diameter ≥ 4 mm. The described Viz.ai Aneurysm CNN performed well in identifying the presence or absence of IAs in an independent validation imaging set. Further studies are necessary to investigate the impact of the software on detection rates in a real-world setting.

A characteristic cerebellar biosignature for bipolar disorder, identified with fully automatic machine learning

BackgroundTranscriptomic profile differences between patients with bipolar disorder and healthy controls can be identified using machine learning and can provide information about the potential role of the cerebellum in the pathogenesis of bipolar disorder.With this aim, user-friendly, fully automated machine learning algorithms can achieve extremely high classification scores and disease-related predictive biosignature identification, in short time frames and scaled down to small datasets. MethodA fully automated machine learning platform, based on the most suitable algorithm selection and relevant set of hyper-parameter values, was applied on a preprocessed transcriptomics dataset, in order to produce a model for biosignature selection and to classify subjects into groups of patients and controls. The parent GEO datasets were originally produced from the cerebellar and parietal lobe tissue of deceased bipolar patients and healthy controls, using Affymetrix Human Gene 1.0 ST Array. ResultsPatients and controls were classified into two separate groups, with no close-to-the-boundary cases, and this classification was based on the cerebellar transcriptomic biosignature of 25 features (genes), with Area Under Curve 0.929 and Average Precision 0.955. The biosignature includes both genes connected before to bipolar disorder, depression, psychosis or epilepsy, as well as genes not linked before with any psychiatric disease. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed participation of 4 identified features in 6 pathways which have also been associated with bipolar disorder. ConclusionAutomated machine learning (AutoML) managed to identify accurately 25 genes that can jointly – in a multivariate–fashion - separate bipolar patients from healthy controls with high predictive power. The discovered features lead to new biological insights. Machine Learning (ML) analysis considers the features in combination (in contrast to standard differential expression analysis), removing both irrelevant as well as redundant markers, and thus, focusing to biological interpretation.

Predicting the Recurrence of Common Bile Duct Stones After ERCP Treatment with Automated Machine Learning Algorithms.

Recurrence of common bile duct stones (CBDs) commonly happens after endoscopic retrograde cholangiopancreatography (ERCP). The clinical prediction models for the recurrence of CBDs after ERCP are lacking. We aim to develop high-performance prediction models for the recurrence of CBDS after ERCP treatment using automated machine learning (AutoML) and to assess the AutoML models versus the traditional regression models. 473 patients with CBDs undergoing ERCP were recruited in the single-center retrospective cohort study. Samples were divided into Training Set (65%) and Validation Set (35%) randomly. Three modeling approaches, including fully automated machine learning (Fully automated), semi-automated machine learning (Semi-automated), and traditional regression were applied to fit prediction models. Models' discrimination, calibration, and clinical benefits were examined. The Shapley additive explanations (SHAP), partial dependence plot (PDP), and SHAP local explanation (SHAPLE) were proposed for the interpretation of the best model. The area under roc curve (AUROC) of semi-automated gradient boost machine (GBM) model was 0.749 in Validation Set, better than the other fully/semi-automated models and the traditional regression models (highest AUROC = 0.736). The calibration and clinical application of AutoML models were adequate. Through the SHAP-PDP-SHAPLE pipeline, the roles of key variables of the semi-automated GBM model were visualized. Lastly, the best model was deployed online for clinical practitioners. The GBM model based on semi-AutoML is an optimal model to predict the recurrence of CBDs after ERCP treatment. In comparison with traditional regressions, AutoML algorithms present significant strengths in modeling, which show promise in future clinical practices.

Comparison of Automated Machine Learning Model Performance for Predicting Chlorophyll-a Concentration according to Measurement Frequency of Input Data

Objectives : Automated machine learning is a recent field of study that automates the process of machine learning model development including proper model selection and optimization. In this study, auto H2O, a novel automated machine learning algorithm, was used to develop a model to predict chlorophyll-a (chl-a).Methods : This study used datasets with different observation frequencies of 1h, 2h, 8h, 24h and 1 week for the development of a machine learning model using an auto H2O algorithm to analyze the effects of measurement frequency of input data on model performance. The effect of the concentration of the input datasets on the performance of the model was also compared by building a model using datasets with observed values of chl-a exceeding 30 mg/m3. The model performance was evaluated using three indices mean absolute error (MAE), Nash-Sutcliffe coefficient of efficiency (NSE) and root mean squared error-observation standard deviation ratio (RSR).Results and Discussion : The MAE, NSE, and RSR of the model using the input data with a measurement frequency of 1h were analyzed as 0.8977, 0.9710, and 0.1704, respectively. The higher the measurement frequency of the input data, the better the performance of the model as the NSE of the model using full data was 0.9710, 0.9552, 0.8856, 0.8396, and 0.7509 for the input datasets with 1h, 2h, 8h, 24h and 1 week observation frequencies, respectively. The difference in model performance according to the difference in measurement frequency was larger for the model using data with the measured value of chl-a exceeding 30 mg/m3, as the NSE was analyzed to be 0.8971, 0.8164, 0.5704, 0.5141, and 0.2052, respectively.Conclusion : The auto H2O model for predicting chl-a showed better model performance as the measurement frequency of the input data increased, and the difference in performance according to the measurement frequency was larger in the range of observed chl-a concentrations that exceeded 30 mg/m3.

A Machine Learning Tool to Monitor and Forecast Results from Testing Products in End-of-Line Systems

The massive industrialization of products in a factory environment requires testing the product at a stage before its exportation to the sales market. For example, the end-of-line tests at Continental Advanced Antenna contribute to the validation of an antenna’s functionality, a product manufactured by this organization. In addition, the storage of information from the testing process allows the data manipulation through automated machine learning algorithms in search of a beneficial contribution. Studies in this area (automatic learning/machine learning) lead to the search and development of tools designed with objectives such as preventing anomalies in the production line, predictive maintenance, product quality assurance, forecast demand, forecasting safety problems, increasing resources, proactive maintenance, resource scalability, reduced production time, and anomaly detection, isolation, and correction. Once applied to the manufacturing environment, these advantages make the EOL system more productive, reliable, and less time-consuming. This way, a tool is proposed that allows the visualization and previous detection of trends associated with faults in the antenna testing system. Furthermore, it focuses on predicting failures at Continental’s EOL.

Automated assessment of the smoothness of retinal layers in optical coherence tomography images using a machine learning algorithm

Quantifying the smoothness of different layers of the retina can potentially be an important and practical biomarker in various pathologic conditions like diabetic retinopathy. The purpose of this study is to develop an automated machine learning algorithm which uses support vector regression method with wavelet kernel and automatically segments two hyperreflective retinal layers (inner plexiform layer (IPL) and outer plexiform layer (OPL)) in 50 optical coherence tomography (OCT) slabs and calculates the smoothness index (SI). The Bland–Altman plots, mean absolute error, root mean square error and signed error calculations revealed a modest discrepancy between the manual approach, used as the ground truth, and the corresponding automated segmentation of IPL/ OPL, as well as SI measurements in OCT slabs. It was concluded that the constructed algorithm may be employed as a reliable, rapid and convenient approach for segmenting IPL/OPL and calculating SI in the appropriate layers.

Automated Cyberbullying Activity Detection using Machine Learning Algorithm

Cyberbullying is the use of technology to harass, intimidate, or harm another person by making hurtful comments, sending threatening messages to humiliate someone in social media. It is important to recognize the signs of cyberbullying activities and takes steps to prevent it. Automated Machine Learning algorithms and Text mining concepts for detecting and classifying bullying messages in social media environment. The abusive texts are clustered using Multinomial Naïve Bayes, LinearSVC, Logistic Regression, K-Nearest neighbour to build a classifier from training datasets. Implementation uses Suspicious-communications-on-social-platforms dataset.

Development of a model to predict water quality using an automated machine learning algorithm

Development of a model to predict water quality using an automated machine learning algorithm

Mechanical response of additively manufactured foam: A machine learning approach

This paper uses ensemble and automated machine learning algorithms to predict the mechanical properties (tensile and flexural strength) of a three-dimensionally printed (3DP) foamed structure. The closed cell foams were made from the most commonly used thermoplastic, High-Density Polyethylene (HDPE). The hollow glass microspheres are infused in HDPE at varying volume %. The available data on these foams' mechanical properties are used by the chosen machine learning (ML) algorithms to propose the best suited algorithm for such a three-phased microstructure as these closed cell foams exhibit. Finally, the strength predictions from the models were validated using experimental data. The models were trained with nozzle temperature, bed temperature, and force values as input parameters. The output parameters predicted were the tensile and flexural strength. LightGBM outperforms all other models in terms of performance among ensemble-based models, while H2OAutoML outperforms all other models. All the ML algorithms produced models with greater than 95% accuracy. Finally, memory and time consumption for each model are presented.